Deep Learning with Differential Gaussian Process Flows
Abstract
We propose a novel deep learning paradigm of differential flows that learn a stochastic differential equation transformations of inputs prior to a standard classification or regression function. The key property of differential Gaussian processes is the warping of inputs through infinitely deep, but infinitesimal, differential fields, that generalise discrete layers into a dynamical system. We demonstrate excellent results as compared to deep Gaussian processes and Bayesian neural networks.
Cite
Text
Hegde et al. "Deep Learning with Differential Gaussian Process Flows." Artificial Intelligence and Statistics, 2019.Markdown
[Hegde et al. "Deep Learning with Differential Gaussian Process Flows." Artificial Intelligence and Statistics, 2019.](https://mlanthology.org/aistats/2019/hegde2019aistats-deep/)BibTeX
@inproceedings{hegde2019aistats-deep,
title = {{Deep Learning with Differential Gaussian Process Flows}},
author = {Hegde, Pashupati and Heinonen, Markus and Lähdesmäki, Harri and Kaski, Samuel},
booktitle = {Artificial Intelligence and Statistics},
year = {2019},
pages = {1812-1821},
volume = {89},
url = {https://mlanthology.org/aistats/2019/hegde2019aistats-deep/}
}